9. Knowledge of Valencian language norms

9.1 Non-existent
9.2 Basic
9.3 Intermediate
9.4 Superior

2.2.The sample

In the face of the impossibility of obtaining data from the entire population, one of the first questions that the researcher must consider is that of designing a sample. The selection is carried out according to the objectives of the research and must be both representative and exhaustive, as the omission of significant sub-groups within the community would have serious consequences.

The starting point must therefore be establishing a sociology known as the real universe of the sample, which is obtained by delimiting the population under study as precisely as possible.

The next step is to determine the number of informants. The problem of the representativeness of the samples has been widely debated in sociological circles, which have produced a whole range of statistical tools to guarantee the reliability of samples. The adoption of these techniques by linguistics has led to a debate that has yet to be definitely concluded. Labov (1966a: 181) defended the principle of homogeneity in linguistic behaviour, which states that linguistic behaviour is more widespread than social behaviour, so that with a small amount of data from a few informants significant conclusions can be drawn. On the other hand, Romaine (1980:91) criticises the use of homogeneity as an absolute and universal principle, and suggests that the meaning of statistics should be checked by objective tests. To take full advantage of the researcher’s work, resources and time, the ability to discover the threshold of meaning, beyond which data becomes redundant, would be ideal. As this calculation is at present impossible, the most common recommendation is to reduce the risk of error by increasing the size of the sample as far as is possible and reasonable (2).

Once the number of informants has been decided upon, it is necessary to select the individuals, applying the sampling technique that is most suitable for the objectives of the research and the situations where it may be carried out. Schematically, we could say that there are two methods of choosing a sample of speakers (Silva-Corvalán 1989:23): a) a survey method in which selection is carried out following random sampling techniques, and b) an intentional selection method. In these methods, either the characteristics of the speakers are predetermined and they are selected more or less at random, or the speakers selected make up a compact group, either because they belong to the same social network, because they live in a common geographical area or because they work together (3).

Having reached this point, the question arises of whether the reproduction of the sample should reproduce the real percentages of the population group. On this subject, Lavandera (1975) says that if the objective is not to discover the general frequency of a linguistic phenomenon within a community, but to establish the correlation between linguistic variation and the population’s extra-linguistic characteristics, a balanced amount of individuals in each category is more useful than a proportional sample.

In the present author’s Doctoral Thesis (Pradilla 1993a), the objective of achieving social stratification of the process of change undergone by the linguistic variable studied makes us decide upon the production of a non-probability sample by quotas (Moreno 1990:123) (4). The selection of informants according to percentages obtained for each factor was carried out using a database created after the tabulation of a survey completed by 200 students at the Batxillerat de Benicarló Institute, in the Baix Maestrat region. Conceived as a sociological exercise that the student had to carry out, the survey supplied details of a total of 1,347 possible informants. The information asked for regarding the student, parents, brothers and sisters and grandparents concerned the social variables under consideration in this study. Once all the informants had been classified, the choice was made at random. It should be mentioned at this point that the initial restriction on placing ourselves within a social network was no longer necessary when we realised that we were dealing with members of a wide range of the Benicarló social spectrum. Finally, to obtain some informants who did not feature in the survey, the snowball or chain technique was used (Noelle 1970:177), which consists of obtaining new informants based on the ones already included.

2.3. Access to informants

Creating a strategy for access to informants that encouraged a receptive and co-operative attitude was without a doubt essential, and would affect the subsequent interview.

We have seen in Pradilla (1993a) how the Batxillerat Institute became a "logistical centre" from where information was selected based on the sociological database. The co-operation of Seminari de Valencià was vital in this respect, as the faculty accreditation was extremely useful in our introduction to students involved in choosing one of their family members from the form they filled in.

After being introduced to the Institute, our next step was to obtain the help of the students within the framework of a generic academic activity, presented as research into the speech of Benicarló. Their role was one of intermediaries responsible for asking their relative to participate in the interview and establishing a relaxed and receptive atmosphere. They were of course told of the methodological requirement of not revealing the linguistic objective of the registration (5). As far as the informants were concerned, the aim of the interview was a social one- "Benicarló as seen by its inhabitants".

Once the helpers had done their job, the researchers’ first contact with the informants took place. This occurred by means of a telephone call to arrange the time and date of the interview.